Two things moved in opposite directions over the past eighteen months. Grid risk went up. Battery capex came down. The point where those two lines cross is where behind-the-meter storage starts to pencil in northern Mexico, and for a large share of industrial loads in Nuevo León, Coahuila, and Chihuahua, they have already crossed.
Take the risk side first. CENACE expects summer 2026 peak demand near 54,000 MW, with a worst-case operating reserve margin around 7%, described by the operator's own director as "apretado," tight (CENACE, reported April 16, 2026). For a plant manager that number is not a policy abstraction. It is the probability that production stops for reasons entirely outside the fence line, on a schedule nobody controls. CENACE declared a state of emergency at 5:04 p.m. on February 18, 2026, and a second at 7:10 p.m., with reserves below 3% against the 6% floor the operator treats as safe and supply interrupted across 21 states. That is what the probability looks like when it lands (Mexico News Daily, February 20, 2026, citing CENACE).
Behind-the-meter battery energy storage (BESS) is best justified in a tight-reserve region as an availability and insurance asset first, not as an arbitrage play. The investment case is the avoided cost of unplanned downtime plus recurring GDMTH demand-charge savings, weighed against a battery capex that fell hard through 2024. Priced that way, storage clears in northern industrial parks that could never justify it on energy arbitrage alone.
What changed on the grid you depend on
The risk went up. The price of the hedge came down.
On risk, the February emergency was not isolated. Mexico ran a comparable alert sequence in May 2024 with reserves near 3% (Bloomberg, Power Technology, May 2024). The structural cause is visible in the build-out numbers. Demand grew 3.4% in 2022 and 3.5% in 2023, against transmission expansion of 0.09% and 0.10% in those years, with CFE grid investment running near 21% of the level PRODESEN recommended (IMCO). Load is climbing in the industrial north while the wires that serve it are close to flat. Thin reserves are the predictable result, and one tripped plant or one bank of cloud is now enough to tip the system.
On price, lithium-ion pack costs dropped roughly 20% across 2024. BloombergNEF put the global average battery pack price at USD 115 per kWh in 2024, the largest annual decline since 2017, with stationary storage among the cheapest end-use segments (BloombergNEF, December 10, 2024). NREL's 2024 Annual Technology Baseline benchmarks a four-hour commercial battery pack at USD 199 per kWh in 2022 dollars, with the pack a minority of the installed system cost once inverter and balance-of-system are added (NREL ATB, 2024). NREL assumes an 85% round-trip efficiency and a 15-year system life for lithium-ion (NREL ATB, 2024). The hardware is now a known, falling quantity. The question is what the avoided loss is worth.
What an outage costs, and what storage saves
Start with the number the finance reader owns. INDEX estimates the cost of a manufacturing interruption at roughly USD 200 million per hour nationally (INDEX). Treat that as an order-of-magnitude industry figure, not a precise reading. Scaled to a single continuous-process site, the per-plant cost of an unplanned outage still runs from tens of thousands of dollars for a clean stop to six figures where it spoils in-process material, damages equipment on an uncontrolled shutdown, or breaks a delivery commitment.
The insurance value has two layers. The first is ride-through. A battery in a behind-the-meter configuration responds in milliseconds and can carry critical load across the seconds-to-minutes voltage sags and interruptions that cause most process trips, without the start delay of rotating backup. The second is bridging. Sized for one to four hours, a battery holds critical load long enough to complete a controlled shutdown or to start longer-duration backup in an orderly way. Neither layer requires the grid to be healthy.
The second value stream owes nothing to grid failure. It is a recurring bill reduction that runs every month the grid behaves. Under CFE's Gran Demanda en Media Tensión Horaria (GDMTH) tariff, the demand-related charges are the capacity charge plus the distribution charge. In the Aguascalientes zone for July 2026 those combine to roughly 475 MXN per kW-month, about USD 27 per kW-month at 17.39 MXN per USD (CFE tariff schedule, July 2026, and DOF reference rate, July 16, 2026). A 2 MW site pays close to 950,000 MXN per month, near USD 55,000, in capacity and distribution charges alone, before a single kWh of energy is billed. That is roughly USD 655,000 per year of pure demand charge (CFE tariff schedule, July 2026).
A battery that discharges into the plant's own peaks trims billed peak demand. Shave 400 kW off the billed peak and the saving is about 400 times USD 27, near USD 10,800 per month, close to USD 130,000 per year on the demand charge alone (derived from CFE GDMTH, July 2026). That saving recurs whether or not the grid ever fails. Stack it on top of the avoided-downtime value and the two revenue lines are additive across the same asset.
The table below shows why the stacked framing changes the answer.
| Value stream | Mechanism | Recurs when grid is healthy? | Anchor figure |
|---|---|---|---|
| Ride-through / avoided downtime | Millisecond response carries critical load through sags and interruptions | No, pays out on grid events | Per-plant outage loss from tens of thousands to six figures per event (scaled from INDEX ~USD 200M/hr national, order-of-magnitude) |
| Peak shaving | Discharge into plant peak cuts billed GDMTH demand | Yes, every month | ~USD 27/kW-month demand charge, ~USD 130,000/yr on a 400 kW shave (CFE GDMTH, July 2026) |
| Power-quality support | Voltage and power-factor support toward Código de Red obligations | Yes | Avoids CRE Código de Red exposure (CRE, grid code, DOF April 2016 and later phasing) |
| Energy arbitrage | Charge off-peak, discharge on-peak on the MEM | Thin in Mexico | Day-ahead average 586.20 MXN/MWh, week of Nov 2-8, 2025 (CENACE weekly report) |
The arbitrage line is deliberately last. In the week of November 2 to 8, 2025, CENACE day-ahead local marginal prices averaged 586.20 MXN per MWh. After the 85% round-trip loss, the off-peak to on-peak spread is too thin to carry a battery on its own (CENACE weekly report, and NREL ATB, 2024). Anchor the case on demand charges and downtime. Treat arbitrage as a rounding error, not a pillar.
What storage costs in 2025
Storage sits inside a policy frame that changed in 2025. The Ley del Sector Eléctrico was published in the DOF on March 18, 2025, with its Reglamento effective October 4, 2025, replacing the 2014 Ley de la Industria Eléctrica. It sets a floor requiring the state to account for at least 54% of energy injected into the grid each year (Greenberg Traurig analysis, February 2025). For a self-supplying industrial, the practical reading is that grid-side relief is capped by design. The state will hold the majority share of injection, and merchant generation that might have added reserves at the margin now enters a narrower door. A plant that wants firmer power has more reason to put the reliability asset on its own side of the meter.
Behind-the-meter storage does not need an injection permit to deliver its core value. Charging from the grid and discharging into on-site load is self-consumption, and the demand-charge and ride-through benefits accrue without selling anything back. That keeps a BESS clear of the interconnection and permitting queue that constrains grid-scale projects.
Sizing in the commercial and industrial segment typically runs from a few hundred kW and several hundred kWh up to the low single-digit MW and MWh range, matched to the critical load a plant must protect and not to total site demand.
Sizing on critical load instead of total demand is the decision that sets the capex here, and the interval data that answers it already sits in your meter. Talk to an advisor.
The risk: when storage does not pay
The case breaks if the grid holds. If summer 2026 clears without a repeat of February, the avoided-downtime line pays nothing, and the battery earns only its demand-charge savings. On a demand-charge-only basis, a system sized primarily for backup carries a longer payback. Backup sizing optimizes for power delivered during an event, not for daily peak-shaving cycles. A plant that buys a large battery purely as insurance and never suffers an outage has bought a policy that did not pay a claim. Price it as insurance, not as an arbitrage return.
Two more risks deserve naming. First, cost trajectory. Pack prices fell in 2024 partly on a fight for market share amid oversupply (BloombergNEF, December 10, 2024). If input costs or trade measures reverse that, installed prices could firm, and a project that pencils today at a falling-cost assumption may not repeat next year. Second, degradation and duration. At 85% round-trip efficiency and roughly one cycle per day over a 15-year design life (NREL ATB, 2024), a battery sized for short bridging will not carry a plant through a multi-hour regional blackout. For long-duration loss of supply, storage is the wrong tool, and pretending otherwise oversells it.
Diesel gensets, honestly compared
The incumbent backup is the diesel genset, and it beats a battery on the one axis that matters in a long outage. Fuel. A genset runs as long as diesel keeps arriving, which a battery cannot.
Where the comparison inverts is speed, stacking, and running cost. A genset takes seconds to start and synchronize, long enough for a sensitive process to have already tripped. A battery bridges the gap in milliseconds, so the two are complements more often than substitutes, the battery covering the ride-through and the genset covering the long tail. On a capex basis, EIA construction-cost data put an internal combustion engine generator near USD 1,248 per kW and a combustion turbine near USD 562 per kW in 2023 (EIA, 2023 data). A genset's headline power cost can look attractive, but it earns nothing on a normal day. It sits idle, consumes maintenance, and carries fuel logistics, emissions permitting, and local air-quality exposure. A battery sized for the same critical load runs its peak-shaving cycle every billing month and offsets part of its own cost against the GDMTH demand charge.
| Dimension | Behind-the-meter BESS | Diesel genset |
|---|---|---|
| Response time | Milliseconds, rides through sags | Seconds to start and synchronize |
| Duration | Limited to stored energy, ~1-4 hr typical C&I | Long, runs while fuel is supplied |
| Recurring value | Peak shaving cuts GDMTH demand charge monthly | None, idle asset between events |
| Fuel and logistics | Charges from grid, no on-site fuel | Diesel supply, storage, and refueling |
| Emissions and permitting | None on site | Local NOx and particulates, air permits |
| Capex basis | Pack ~USD 115/kWh (BNEF, 2024), 4-hr pack ~USD 199/kWh, 2022 USD (NREL ATB) | ICE ~USD 1,248/kW, CT ~USD 562/kW (EIA, 2023) |
| Round-trip / efficiency | 85% round-trip (NREL ATB, 2024) | Continuous fuel burn while running |
On our read, the right answer in a tight-reserve northern park is rarely one or the other. It is a battery sized for instantaneous ride-through and daily peak shaving, backed by rotating generation for the rare long-duration event, with the battery's recurring demand-charge savings doing the work of shortening the combined payback.
How to size and justify the system
Do not buy a battery to beat the CENACE price on the mercado de balance. That case does not clear at current spreads. Buy it to protect the loads that cannot tolerate an uncontrolled stop and to trim the GDMTH demand charge you already pay every month.
Three moves, in order.
Size on critical load, not on total demand. Identify the specific lines whose uncontrolled shutdown destroys material, damages equipment, or breaks a delivery, and size the battery to bridge those loads for the minutes needed to complete a controlled stop or start rotating backup.
Underwrite on two lines, not one. Put the recurring GDMTH demand-charge saving in the base case, near USD 130,000 per year on a 400 kW shave at the July 2026 Aguascalientes rate. That is the line that recurs every month. Treat avoided downtime as event-driven upside, carrying a probability tied to the roughly 7% summer reserve margin. If the demand-charge line alone gets the payback inside the equipment life, the insurance value is a free option on top.
Move before the summer peak, not after it. The reserve margin CENACE flagged for summer 2026 is a forecast the operator has already published, and pack prices are at a record low BloombergNEF called the largest annual drop since 2017 (CENACE, April 16, 2026, and BloombergNEF, December 10, 2024). Procurement, sizing studies, and interconnection notification for a behind-the-meter system take months. A plant that starts specifying after the next emergency has already taken the loss the asset was meant to prevent.
The number a CFO can carry into the budget meeting is this. A 2 MW site pays near USD 655,000 a year in GDMTH capacity and distribution charges. A right-sized battery that shaves 400 kW of billed peak returns on the order of USD 130,000 a year, before a single grid event. The avoided-downtime value sits on top of that, measured against a summer the grid operator itself calls tight.
Talk to an advisor
Solar and BESS Feasibility Study
Mexico Energy Partners sizes the battery against the loads that cannot take an uncontrolled stop, models the recurring GDMTH demand-charge line separately from the event-driven downtime line, and compares both against a diesel genset. Send twelve months of CFE bills and meter interval data, and the value-stack model comes back in time to act before the summer peak.
Talk to a Mexico Energy Partners advisor or email info@mexicoenergypartners.com. Mexico Energy Partners sells no equipment and is compensated only by the client.
Sources
- Mexico News Daily, February 20, 2026, citing CENACE, on the February 18, 2026 states of emergency and reserves below 3%.
- CENACE, reported April 16, 2026, on summer 2026 peak demand near 54,000 MW and a worst-case operating reserve margin around 7%.
- Bloomberg / Power Technology, May 2024, on the May 2024 grid alert sequence with reserves near 3%.
- IMCO, on 2022-2023 demand growth (3.4% and 3.5%), transmission expansion (0.09% and 0.10%), and CFE grid investment near 21% of PRODESEN-recommended level.
- INDEX, manufacturing interruption cost estimate of roughly USD 200 million per hour nationally, order-of-magnitude.
- CFE tariff schedule, July 2026, GDMTH capacity and distribution charges, Aguascalientes zone.
- DOF reference exchange rate, 17.39 MXN per USD, July 16, 2026.
- BloombergNEF, December 10, 2024, average lithium-ion battery pack price of USD 115 per kWh in 2024.
- NREL Annual Technology Baseline, 2024, commercial battery storage four-hour pack cost USD 199 per kWh (2022 USD), 85% round-trip efficiency, 15-year life.
- EIA construction-cost data, 2023, internal combustion engine generator USD 1,248 per kW and combustion turbine USD 562 per kW.
- CENACE weekly report, week of November 2-8, 2025, day-ahead average local marginal price 586.20 MXN per MWh.
- Ley del Sector Eléctrico, DOF March 18, 2025, and Reglamento effective October 4, 2025, with a state 54% injection floor per Greenberg Traurig analysis, February 2025.
- CRE, Código de Red (Disposiciones administrativas de carácter general), DOF April 2016 and later phasing.